On Sun, Feb 22, 2015 at 3:38 PM, Sebastian Huber <sebastian.hu...@embedded-brains.de> wrote: > The start.S already has a #include <bspopts.h>. So I would add a BSP option > to define something like BSP_START_DO_LOCKSTEP_INITIALIZATION and use it in > start.S.
Thanks for your answer, we'll look into this. >> >> Additionally, when the CCM detects an error it raises a FIQ which must >> be handled in a board-specific way, otherwise it re-raises the FIQ as >> soon as _ARMV4_Exception_fiq_default enables FIQs in the CPSR. > > > Oh, this is a bug, the _ARMV4_Exception_fiq_default should not enable the > FIQs. It's actually in save_more_context, at the point where it's loading the SPSR plus the T and I bits into the CPSR. If we got there from the FIQ handler, SPSR would have FIQs enabled. > I would add BSP_START_EXCEPTION_UNDEF_HANDLER etc. defines and add something > like this: > > #ifdef BSP_START_EXCEPTION_UNDEF_HANDLER > .word BSP_START_EXCEPTION_UNDEF_HANDLER > #else > .word _ARMV4_Exception_undef_default > #endif You mean at the beginning of start.S? If it's OK with you then we'll get to work on that. Again, thanks a lot for your answer. _______________________________________________ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
